SKAM-ResistNet — TEM1 Target Finder

Live Demo: https://6135d67466b5563897.gradio.live/

---

Project Description

SKAM-ResistNet is a lightweight deep learning application for predicting small-molecule binding affinity to TEM-1 β-lactamase, an enzyme responsible for antibiotic resistance in bacteria.

The tool:

• Takes a set of chemical compounds in SMILES format.
• Predicts binding affinity (pAff, −log10 Kd in molar).
• Outputs a calibrated binder probability.
• Generates visual plots to interpret predictions.

This project was developed for the HackNation Global AI Hackathon by Team SKAM.

---

Features

• Accepts user-provided SMILES strings.
• Predicts binding affinity to TEM-1 β-lactamase.
• Provides binder probability with confidence intervals.
• Generates bar, scatter, and heatmap visualizations.
• Lightweight, fast, and accessible for small-scale labs.

---

Setup Instructions

1) Clone this repository

2) Install dependencies (requirements.txt)

3) (Optional) Add dataset

4) Run the application

---

Dependencies

Core packages used:

• pandas
• numpy
• matplotlib
• torch
• gradio
• scikit-learn
• xgboost
• transformers See requirements.txt for the full list.

---

Dataset
BindingDB
Google Drive Dataset Link

Team SKAM

• Samhitha Kunadharaju
• Aditi Mod